Age | Commit message (Collapse) | Author |
|
The trick we use to allow this is undefined behaviour and is not
liked by LLVM. We should stop using it but it has the potential to
break to many things for a minor release.
|
|
- warning: absolute value function 'abs' given an argument of type 'long' but has parameter of type 'int' which may cause truncation of value [-Wabsolute-value]
- warning: 'this' pointer cannot be null in well-defined C++ code; pointer may be assumed to always convert to true [-Wundefined-bool-conversion]
|
|
|
|
|
|
Options enabled:
- SSL_OP_NO_SSLv2
- SSL_OP_SINGLE_DH_USE
- SSL_OP_SINGLE_ECDH_USE (if it exists)
Partial backport of #856 by @jvehent
|
|
|
|
Fix hash_map.h on non-GNU C++ standard library implementations.
|
|
Revert "Fix an off-by-one on registration timeout handling."
|
|
|
|
This seemingly breaks (occasionally timing users out a second early), possibly
due to rounding issues on time (User::age and Time() both return int, but a
whole second is comprised of many ms).
The root cause needs investigation, but we cannot leave a potentially breaking
bug in the tree.
This reverts commit 4414d644a163f3906f90b35186e07ce0383161f4.
|
|
Fix an off-by-one on registration timeout handling.
|
|
The use of > instead of >= here added a second onto the configured timeout.
We won't be breaking any existing configuration, because users.h explicitly
increases the limit if it's 0.
|
|
struct is reallocated
|
|
|
|
parameter but does not have one
Thanks to @docwhat for helping to track this down, fixes issue #924
|
|
This code is incorrect and can produce addresses with multiple double
colons. No other IRCds do this and most systems today will give us the
compact version anyway.
Issue #914
|
|
different from the type of the query. #66
|
|
|
|
Fix unit-cc not loading configure cache when building files.
|
|
|
|
[2.0] Fix various problems with the build system.
|
|
The extend_pkg_path method will be removed in 2.2 as it
causes problems with non-standard pkg-config paths. This
commit adds a workaround until that happens.
|
|
|
|
[2.0] Fix listmodes when the config does not specify a wildcard size entry.
|
|
|
|
|
|
Mention m_regex_stdlib in the list of regex providers after m_filter
|
|
collision if the nick is used by a (local) client that isn't fully registered
|
|
|
|
|
|
|
|
Fix various small problems with configure.
|
|
- Fix temporary files not being deleted after use.
- Fix the option for disabling rpath.
- Fix the help message:
* Fix indentation of some help options.
* Fix default paths not containing 'run'.
* Remove documentation for --disable-ipv6 (exists but does nothing).
* Remove documentation for --library-dir (does not exist).
- Fix some minor errors in messages.
|
|
Fix copy/paste error in m_denychans.
|
|
|
|
SASL)
|
|
Remove whitespace from docs
|
|
|
|
|
|
new nick param rewritten to the uuid
Suppressing these messages meant that we relied on an appropriate NICK to
come from the direction the SAVE was sent to; this left all servers behind
us uninformed until the target server of the SAVE reacted with a nick change
to uuid.
This was problematic because someone can legitimately change nick to the
past nick of the SAVEd user on a server that already has the SAVEd user
with the uuid nick and that nick change message can reach servers that
haven't yet seen the nick change to uuid of the SAVEd user.
|
|
provides an IOHook
|
|
Rewrote "invalid credentials" error message to remove confusion regarding "server snomask", which some users interpreted as "I must make sure the snomask on each server is identical" (which makes no sense), when really the intention is basically "look at the logs on the other server".
|
|
Replace deprecated entry in launchd config with the modern equivalent.
|
|
|
|
|
|
The call to remove() can fail if the database doesn't already exist. Just
ignore the error, which effectively makes the rename() call below work as
it does on nix
|
|
Their functions are all inlined and Windows gives unresolved symbol errors
|
|
OnCheckReady()
|
|
|
|
the source of the FJOIN is bursting or not
Previously the flag was true if our route to the source was bursting, which resulted in m_joinflood incorrectly registering netjoins when a remote server was bursting
Fixes issue #839 reported by @joshtek0
|